Where do people come up with OW statistics and numbers from?

98% of the threads on here revolve around nerfs or buffs to heroes and I see people constantly pull numbers out of a hat regarding:

*Pickrates
*GM Pickrates
*OWL picks

People post their for or against arguments with confidence by citing these pickrates. The problem is, if you use the Search function, every hero has been called out as having a low pickrate or a high pickrate. Meaning:

You can read one thread where people are talking about, lets say Sombra, and saying how she has a 70% high pickrate and then another thread calling for Sombra to get buffs because she has a less then 20% pickrate. (I made this up as an example however, seriously search the word “pickrate” and look at the threads…).

Nearly every single hero here is a high and low pick rate simultaneously. Talk about fake news. Where do these numbers regarding “pickrates” actually come from and how faithful/true are they?

There is no way in hell the majority of heroes can all have high pickrates and low ones, at the same time. So which is it? Why should anyone believe anything about pickrates?

General Playerbase, Sombra has a 1.25% pickrate, Top 1%, she has a 2.15% pickrate.

Overall, the highest picked Tank is still Reinhardt at 8.6%, but in GrandMasters he’s at 2.97% which makes him the fifth popular tank among the high tiers players.

There’s is no 70% pickrate, the highest you can go is around 16%. Numbers that high usually refer to someone’s personal win rates.

Which part of the population do you relate to? The general playerbase? The top 1%? or the Pro?

usually a stat site like overbuff.

1 Like

Almost all stats come from overbuff (website). It used to be a good tool for analyzing stats but it’s not quite as useful these days due to private profiles. Data from private profiles isn’t available to stat sites.

I believe overbuff and overwatch.op.gg are the most often consulted sites.

You can find both good and bad numbers for some heroes by being selective about the time range and/or rank range that you refer to.

When people talk about using a site to track their personal stats, they nearly always refer to overbuff, so I suspect it is tracking a larger portion of the player base, but that’s just my impression.

I’m sure both of them are tracking a statistically significant portion of the player base, but it is a mostly self-selected group (people with public profiles who have either checked their stats on these sites, or who have had someone else check their stats on these sites). It probably skews a little towards the ‘tryhards’ since tracking your own stats suggests that you’re trying to improve. That might skew all the winrates a little bit high, but I can’t think of any other bias it would introduce.

Probably sites like Overbuff which lost a lot of their actual accuracy since private Profiles were introduced as they don’t get any info straight from Blzzard’s Data Bases.

For general ladder stats, overbuff.com it has a sample size of about 500,000 players, which gives a lot more than a 99% confidence interval for information for the general population (which would require 16,500 profiles). Most polling uses a 95% confidence interval, which would require fewer than 1,000. So it’s still mathematically, a very, very good source of data.

Since things went private by default, there was basically no shift in the patterns of data.

For OWL stats, winstonslab.com has everything you’ll ever need, and more.

1 Like

Over 70% pick rate is used from other sites, that calculate it differently than overbuff. Mainly seen with pro pick rates, where some1 like lucio is picked 90% of the time by a team.

There is also the fact that pick-rates cannot be uniformly distributed.

It’s simply impossible for a game like this.

And things can affect the pick-rate and win-rate of a hero without the hero being under-powered.

All of these threads effectively commit a black and white fallacy by basically saying their interpretation of the statistics is true while disregarding every other possible explanation without reason.

This is what happens when you have mathematically and statistically ignorant people trying to come to conclusions about something as dynamic as OW balance.

Can you think of an example when overbuff was way off the mark about a hero?

It has always been very good at reflecting what is happening in game

Yep. When profiles became private, there was no other change around that time so performances didn’t change. The difference from before and after privatization was an improvement of around 1% or less. That’s barely any change.

If anything, actual server numbers would be worse. The people making up Sombra on in Overbuff are the ones actually trying. Include the rest of the playerbase, and Sombra’s win rate would be even lower.